How is learning triggered?

[Michael Sherwin]

In engine - engine games learning is triggered with the result command. I just assumed that in human - engine games a result command would be sent when the human resigns, but, this does not seem to be the case. What guis' send a result command in human - engine games and when or is there another way to trigger learning? THANKS!

[George Lyapko]

What GUI do you talk about?

Winboard sends "result" command in human-engine game when human resigns or loose on time:

Code: Select all

1078 <first : Bestia v0.88b2 (c) 2000-2002 by George Lyapko
1078 <first : hash size=12582900
1078 <first : pawn hash size=3145704
1078 <first : log=TRUE
1078 <first :
1078 <first : feature analyze=1 setboard=1 ping=1 name=1 done=1
1078 >first : accepted analyze
1078 >first : accepted setboard
1078 >first : accepted ping
1078 >first : accepted name
1078 >first : accepted done
1094 >first : new
random
1094 >first : level 0 1 0
1094 >first : post
1094 >first : hard
1094 >first : easy
1094 >first : ping 1
1109 <first : pong 1
3750 >first : name lyapkyur
3750 >first : time 6000
otim 6000
3750 >first : e2e4
4375 <first : inbook
4375 <first :   0     0         0          0(book move) c7c5
4375 <first : move c7c5
Interrupting first
5734 >first : time 5937
otim 5864
5734 >first : f2f3
6234 <first : inbook
6234 <first : best=3 secondbest=-5
6234 <first : ponder=FALSE
6234 <first : Static eval=-59
6234 <first : -Failing low
6234 <first :   2   -22        50        125 d7d5 d2d3
6234 <first :   2   -43        50        232 d7d5 d2d4
6234 <first : +Failing high
6234 <first :   3   -18        50        913 e7e6
6234 <first :   3     4        50       1679 e7e5 b1c3 d7d6
6234 <first : -Failing low
6234 <first :   4   -21        50       4040 e7e5 b1c3
6250 <first :   4   -34        50       6881 e7e5 b1c3 d8h4 g2g3 h4h5
6250 <first :   5    -9        52       9276 e7e5
6250 <first : +Failing high
6250 <first :   5    -9        52       9276 e7e5
6250 <first :   5    10        52      13496 e7e5 b1c3 g8f6 f1b5 b8c6
6328 <first :   5    10        59      23724 e7e5 b1c3 g8f6 f1b5 b8c6
6344 <first :   6   -15        59      27530 e7e5 b1c3
6531 <first : -Failing low
6531 <first :   6   -15        80      70960 e7e5 b1c3
6609 <first :   6   -34        88      89644 e7e5 b1c3 g8f6 f1b5 b8c6 d2d3
6703 <first :   6   -33        97     117497 d7d5
6875 <first :   6   -27       114     150210 d7d5 f1b5 c8d7 b5d7 b8d7 e4d5 d7b6 d2d4
6969 <first :   6   -27       123     179903 d7d5 f1b5 c8d7 b5d7 b8d7 e4d5 d7b6 d2d4
6969 <first : move d7d5
GameEnds(32, White resigns, 2)
Interrupting first
8859 >first : result 0-1 {White resigns}


[Michael Sherwin]

Hi George,

I was using Arena 1.08 and noticed that no result command was sent. And it did not show in the debug window either. I just did not want to use the 'A' word in my original post. This is good news though, as I can tell human players that they can at least use Winboard if they wish to use Romi's learning feature. Hopefully there are other interfaces that will work as well.

THANKS,
Mike